340 VON AKX [CHAP. 16 



to make optical measurements of the zenith angle of the horizon rather un- 

 certain and, possibly, of dubious value for the measurement of regional slopes, 

 except when the air-sea temperature contrast is almost nil. 



As a means for studying the changing behavior of the horizon, a dipmeter 

 was built, as shown in Fig. 13a, b, to make optical measurements of the zenith 

 angle of the horizon from land-based stations. Basically this device consists of a 

 good-quality 5-in. refracting telescope fitted with a reticle eyepiece against 

 which the altitude of the visible horizon is measured. The telescope is leveled by 

 first blocking the view of the horizon with a shutter beyond the pentaprism 

 and then illuminating the reticle from the eye end so as to produce a collimated 

 beam of light. The collimated beam enters a pentaprism and is reflected back 

 upon itself from a quiet mercury surface housed in the cylindrical compartment 

 between the tripod legs. When properly focused, both the reticle and its image 

 can be clearly seen at the same time and made to coincide through suitable 

 rotations of the alignment screws. 



In principle, the reticle and its image define a line normal to the mercury 

 surface. If the reticle and its image are made to coincide before each observation 

 of the horizon on a new azimuth, the apparent departure of the horizon from 

 90° zenith angle can be measured directly. 



But the pentaprism deviation differs slightly from 90°. This error introduces 

 a small systematic angular elevation or depression of the optical axis of the 

 telescope from horizontal. Owing to dip and the prism error, the regional 

 inclination of the circle of the horizon with respect to local vertical can be 

 obtained only by plotting the apparent zenith angle of the horizon as a function 

 of azimuth relative to the center of a polar diagram. If the horizon figure lies 

 some 90° from the local vertical, the apparent zenith will be uniformly separated 

 from the center of the polar diagram on all azimuths and a circle of best fit 

 will have its center at the origin, as shown in Fig. 14. In the presence of a 

 regional slope, however, the center of the circle of best fit will depart from the 

 center of the polar diagram by an amount, r, proportional to the angle between 

 the local vertical and the normal to the horizon plane. The actual patterns of 

 points obtained from any station may not be circular owing to real or apparent 

 buckling of the horizon, as, for example, when the azimuth of view swings 

 across the backs of waves and into their troughs. The assumptions made in this 

 treatment of the observations are that the hydrostatic effects of barometric 

 loading can be measured and eliminated, and that the direction of gravity 

 within the circle of the horizon is conically symmetrical around the point from 

 which the observations are made. It is also assumed that the effects of atmo- 

 spheric refraction are uniform on all azimuths. This last is, perhaps, the weakest 

 point of the method except for the fact that, at sea, the lapse rate of temperature 

 through the sea surface into the lower layers of the air can be remarkably 

 uniform over large areas. 



In good weather, when the temperature difference between the air and the 

 sea is not greater than 1°C and the winds are so light as to produce only small 

 ripples (when the sea is glassy smooth the horizon is invisible), the horizon can 



